Overload Atlas

Inspiration

1 in 5 college students experience sensory processing challenges. Yet zero tools exist to help them preview campus environments before experiencing overwhelming situations. I've watched neurodivergent students struggle with daily uncertainty:

Will the cafeteria trigger a meltdown today? Where can I find a quiet study spot? Which classroom route has the least fluorescent lighting?

Up to 80% of individuals with autism experience sensory processing differences. They deserve better than trial-and-error navigation that costs them energy, focus, and mental health. This inspired Overload Atlas - a tool to preview sensory conditions and prepare with AI-guided strategies, transforming anxiety into informed confidence.

What it does

Overload Atlas is an immersive 3D sensory simulator that lets students experience NJIT campus locations before visiting them in real life. Navigate 4 campus environments:

Quiet Library Study Room (low sensory sanctuary) Campus Center Cafeteria (high-intensity chaos) Standard Lecture Hall (moderate balanced space) Busy ECEC Hallway (fluorescent flicker challenge)

Each location has 3 distinct zones with unique sensory profiles. As you explore in first-person view, you receive: ✅ Live sensory metrics - Noise, light, and crowd levels (1-5 scale) ✅ Real-time Claude AI guidance - Three types of advice:

Analysis: Current environment assessment Quick Insights: Research-backed statistics Survival Strategies: Concrete coping mechanisms

✅ Interactive HUD - Stress meter, timer, mini-map, zone detection ✅ Immersive effects - Visual representations of sensory overload (strobing lights in ECEC, pulsing chaos in cafeteria) Example AI guidance (Cafeteria Main Hall):

"Peak chaos: 200+ simultaneous conversations. Studies show this environment triggers fight-or-flight in 60% of autistic individuals. Strategy: Noise-canceling headphones + tinted glasses. 15-minute max exposure."

Students can explore, learn their limits, and develop personalized navigation strategies - all without risking real-world meltdowns.

How I built it

Frontend - Three.js 3D Engine:

Custom first-person controller with WASD movement Quaternion-based camera rotation (no gimbal lock!) AABB collision detection for walls and furniture Dynamic lighting effects per environment

AI Integration - Claude Sonnet 4.5:

Real-time contextual advice generation Zone-specific guidance (12 unique zones across 4 locations) Research-backed recommendations with statistics Three-tier advice system (Analysis + Insights + Strategies)

UX Design:

Mini-map with live player position and direction indicator Stress meter calculating cumulative sensory load Timer tracking exposure time per room Multiple HUD panels with Claude AI branding

Technical Highlights:

Zero dependencies (vanilla JS + Three.js CDN) Pointer Lock API for true first-person control Canvas-based mini-map rendering Modular room data structure for easy expansion

Built in under 10 hours at the NJIT Claude Builder Club Hackathon.

Challenges I ran into

1. First-Person Controls Initially, movement was inverted and camera rotation broke in certain rooms. Fixed by:

Properly implementing Euler angles with YXZ order Using quaternions for smooth rotation Applying direction vectors relative to camera orientation

1. Collision Detection Players could walk through walls and furniture. Implemented AABB (Axis-Aligned Bounding Box) collision with separate X/Z axis checks.
2. Sensory Overload Representation How do you show sensory overload in 3D? Solutions:

Strobing background colors for fluorescent flicker Pulsing effects for chaotic environments Color psychology (red = danger, green = calm)

1. Balancing Information Density Too much text = overwhelming (ironic for an anti-overload tool!). Solved with three separate Claude AI windows, each focused on one aspect.
2. Performance Multiple animated elements could cause lag. Optimized with:

Minimal polygon counts Efficient update loops (500ms intervals vs. every frame) Reusing geometry for similar objects

Accomplishments that we're proud of ✨ Built a fully functional 3D simulator in <10 hours ✨ 12 unique zones with distinct AI-generated guidance ✨ Research-backed sensory ratings (decibel levels, cortisol studies, migraine triggers) ✨ Smooth first-person controls that actually work on trackpad and mouse ✨ Zero build process - works instantly in any browser ✨ Genuine accessibility impact - solves a real problem for a real community

Accomplishments that we're proud of

Most proud of: The Claude AI integration feels genuinely helpful, not gimmicky. Each piece of advice is specific, actionable, and backed by research. This isn't "AI for AI's sake" - it's AI meaningfully improving lives.

What we learned

Technical:

Quaternions are superior to Euler angles for 3D rotation Pointer Lock API subtleties across browsers Collision detection performance trade-offs Canvas rendering for 2D overlays on 3D scenes

Design:

Less is more - three focused advice panels beat one cluttered wall of text Color psychology matters deeply for sensory-sensitive users Visual effects can show sensory concepts that words can't capture

Impact:

Accessibility isn't an add-on feature - it should be the core product AI excels at contextual, personalized guidance (not generic responses) Neurodivergent users need preparation tools, not just accommodations

Biggest lesson: Building with the community you're serving is essential. The sensory ratings, advice content, and survival strategies reflect real neurodivergent experiences - not assumptions.

What's next for Overload Atlas

Short-term:

Expand to all NJIT campus buildings (currently 4 locations) Add user-submitted sensory reports (crowdsourcing) Time-of-day simulation (morning vs. afternoon crowds)

Medium-term:

Mobile app for on-the-go sensory checking Integration with campus maps and class schedules Partnership with NJIT Disability Services for official adoption

Long-term:

VR headset support for full immersion Expand to other universities (template system for new campuses) Real-world sensor integration (actual decibel readings via IoT devices) Personalized profiles (my sensory thresholds vs. yours)

Vision: Every neurodivergent student deserves to navigate campus with confidence, not anxiety. Overload Atlas is the first step toward making sensory accessibility as standard as wheelchair ramps.

Built With

• canvas-api
• claude-ai
• html5
• javascript
• three-js
• webgl